Main idea is that writing phonetic text by using "combining characters" to put phonetic accents on characters is braking spell check because special characters are added to correctly spelled words. Suggestion is to add option to filter/ignore combining characters in spell checking preferences. This is a problem especially for cyrillic script because unicode does not have (for example) vowels with acute like it does for latin script. That is why one must use combining characters (for example U+0300 Combinig grave accent, U+0301 Combining acute accent, ...) to accentuate characters used mainly for pronunciation purposes. Example (Russian): accentuated non-accentuated пожа́луйста пожалуйста спаси́бо спасибо Example (Serbian): accentuated non-accentuated м̋олим ва́с молим вас хв̂ала ва́м хвала вам (copy this text to Writer and set the appropriate language with automatic spelling turned on; note: you must install bundled extensions for spell check for appropriate language) Another example (in Serbian) using accents to distinguish homonyms: Ја са́м с̂ам. - I am alone.
Adding the next line to the aff file of the Hunspell dictionary, it's possible to ignore all diacritics during spell checking: IGNORE ̀́̃ (i.e. IGNORE U+0300 U+0301 U+0302)